{"id":"d7fa8ef8-c61c-4be3-8b66-7d93a84251de","arxiv_id":"2412.14607","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"A 3.9-sigma dipole in the Tully-Fisher zeropoint could indicate 3% anisotropic Hubble expansion, but it is degenerate with a bulk flow; future surveys should resolve it.","lead":"This paper fits a direction-dependent Hubble constant to Tully-Fisher galaxy distances and finds a 3.9-sigma dipole that could be a 3% variation in H0, or just a large-scale bulk flow. It then forecasts that upcoming WALLABY and DESI surveys will be able to tell the two apart and detect a 1% anisotropy at 5.8 sigma.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Dipole robustness hinges on unmodeled H I zero-point offsets: the best-fit dipole is dominated by a north-south signal that a 0.01 dex calibration offset could fully mimic.","rationale":"The reader's weakest_assumption correctly identifies the spatial uniformity of the Tully-Fisher relation and the heterogeneous H I line-width compilation as the load-bearing point. My read confirms this and sharpens it: the dipole's dominant z-component makes it especially sensitive to a north-south zero-point offset, and the paper's own quoted 0.01 dex threshold is exactly the size of the claimed effect. The paper is otherwise careful: it discloses the bulk-flow degeneracy (Table 1 gives ln B = 4.7 favoring a velocity dipole), presents honest Bayes factors, and constructs conservative forecasts. The absence of an explicit systematic error budget for H I line-width calibrations is the main gap. Since the paper is already CONDITIONAL and this concern does not overturn the central claim but does reinforce the need for a systematic check, the reader's verdict stands unchanged.","tokens_in":14465,"tokens_out":3284,"duration_ms":31515,"concrete_test":"Re-fit the full forward model with an additional free zero-point offset per H I source catalog (e.g., ALFALFA, ADHI, Springob/Cornell, Pre Digital), and examine the marginal posterior of the a0 dipole and the offset values. If the dipole amplitude shifts below ~2σ or the inferred offsets are consistent with 0.01 dex, the central claim is not robust. As a complementary check, split the sample by hemisphere and fit the dipole separately to northern-only and southern-only galaxies; a significant difference in amplitude or direction would corroborate a calibration-induced signal.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The 3.9σ dipole claim rests on the assumption (Section 3) that variations in the Tully-Fisher zeropoint a0 are due only to H0 variations, not to position-dependent systematics. The data, however, combine H I line widths from heterogeneous sources: ALFALFA covers only the north, while southern sources come from ADHI, Springob/Cornell, and the Pre Digital catalog (Section 2.2). The best-fit dipole is predominantly a z-component: a0z = 0.048 ± 0.014 mag out of total amplitude 0.063 ± 0.016 mag, i.e., mostly a north-south anisotropy. The paper itself notes that a systematic difference as small as Δlog W_mx^c = 0.01 corresponds to roughly a 3% effect on H0 (Section 2.2), which is the same size as the claimed dipole (ΔH0/H0 ≈ 3%). The KS test (p = 0.077) only compares full line-width distributions; it does not constrain the zero-point calibration offset between northern and southern samples, nor does the paper fit any per-source zero-point nuisance parameters. Therefore a modest, plausible H I calibration offset could create the entire observed dipole, and the quoted 3.9σ is a purely statistical significance that does not include this systematic.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper uses the Tully-Fisher relation in the Cosmicflows-4 catalog to search for directional variation in the Hubble constant. The authors fit a zeropoint monopole, dipole, and quadrupole to the W1-band TF data, report a 3.9σ dipole of amplitude 0.063±0.016 mag in the direction (ℓ,b)=(142±30°,52±10°), and translate this into a possible 3% H0 variation. They compare the H0 dipole model with a constant H0 plus bulk flow model using BIC, finding the bulk flow model strongly favored. Using injection-recovery mocks, they then forecast that future WALLABY and DESI TF data could detect a 1% H0 dipole at 5.8σ and distinguish it from a ΛCDM bulk flow.","tokens_in":14780,"tokens_out":16060,"duration_ms":97332,"significance":"The paper offers a plausible new avenue for testing the cosmological principle with TF data, and the careful model comparison between an H0 dipole and a bulk flow is a useful step. The authors are transparent about the degeneracy between these interpretations. The mock-based forecasts for WALLABY/DESI are timely and the injection-recovery procedure is a strength. However, the headline dipole significance and the forecast detection claim are undercut by (i) an unmodeled north-south H I calibration systematic that the paper itself notes is comparable in size to the signal, and (ii) a numerical inconsistency in the forecast significances. These issues need to be resolved before the central claims can be accepted.","major_comments":[{"comment":"The statement in the abstract that 'A model that includes this H0 dipole is only weakly favored relative to a model with a constant H0 and a bulk motion' is contradicted by Table 1, where the velocity dipole (bulk flow) is the preferred model with lnB01 = 4.7, which the authors themselves call 'strong evidence' in Section 4.2. As written, the abstract reverses the direction of the model comparison and could mislead readers into thinking the H0 dipole is preferred over the bulk flow. The abstract should state that the bulk-flow model is favored over the H0 dipole model, or that the H0 dipole is not preferred once a bulk flow is allowed.","section":"Abstract and Section 4.2, Table 1"},{"comment":"The quoted 3.9σ significance of the dipole is purely statistical and does not include the dominant systematic from heterogeneous H I line-width sources. The paper notes in Section 2.2 that a 0.01 dex offset in log W corresponds to a 3% effect on H0, the same size as the claimed dipole, and the best-fit dipole is dominated by a0z = 0.048 ± 0.014 mag, i.e., a north-south asymmetry. The KS test (p = 0.077) cannot constrain zero-point offsets between the northern ALFALFA-based sample and the southern (ADHI/Springob/Cornell/Pre-Digital) samples, and no per-catalog zero-point nuisance parameters are included in the fit. A plausible calibration offset could therefore produce the entire dipole. The authors should fit per-catalog zero-point offsets or otherwise demonstrate robustness to 0.01 dex offsets, and propagate this systematic into the reported significance.","section":"Section 2.2 and Section 4.1, Eqs. (3.5)-(3.7)"},{"comment":"The forecast that WALLABY+DESI will detect a 1% H0 dipole at 5.8σ is not supported by the presented numbers. The text states that a 3% H0 dipole is detected at 9σ when fitting for the H0 dipole; under Gaussian scaling, a 1% dipole should then be detected at about 3σ, not 5.8σ. The authors do not show the significance distribution for the 1% injection, nor do the mocks include the north-south line-width calibration uncertainty that the paper itself identifies as a persistent limitation even for future datasets. Either provide the full injection-recovery results for the 1% case and reconcile the scaling, or soften the 5.8σ claim to reflect that it is conditional on the absence of calibration systematics.","section":"Section 5.2"}],"minor_comments":[{"comment":"The abstract contains a typo: 'However, m simulations that the expected Tully-Fisher data...' appears to be missing a verb or phrase; it likely should read 'our simulations show that'.","section":"Abstract"},{"comment":"The KS test p-value of 0.077 is marginal; the phrase 'providing no evidence the distributions differ significantly' should be softened to 'providing no strong evidence'.","section":"Section 2.2"},{"comment":"There is a typo in the sentence 'a bulk flow has a an error of only ± 10 km/s in each component'.","section":"Section 5.1"},{"comment":"In the conclusions, 'the uncertainties in measurements H0' is missing the word 'of' before 'H0'.","section":"Section 6"},{"comment":"The forecast combines WALLABY and DESI but does not include FASHI, despite FASHI being introduced earlier as a future H I survey with more than 100,000 sources; a sentence justifying the exclusion would strengthen the forecast.","section":"Section 5.2"}],"recommendation":"major_revision","confidential_remarks":"The abstract's reversal of the model-comparison result is a serious issue and should be fixed at revision stage. The forecast significance scaling (3% giving 9σ versus 1% giving 5.8σ) suggests a possible error in the mock analysis; the authors should be asked to provide the full distributions and clarify the procedure. The systematic calibration issue is the most serious scientific concern and should be addressed with per-catalog zero-point fits or an explicit systematic budget before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a solid, well-qualified paper that applies a forward-modeling Tully-Fisher pipeline to a new question. The headline 3.9σ H0 dipole is honestly presented as degenerate with a bulk flow. The real soft spot is the unmodeled north-south H I calibration offset, which could produce the entire signal. That doesn't kill the paper, but it needs to be addressed before the claim is taken at face value.\n\nWhat's genuinely new: using differential TF zeropoint variations from CF4 to constrain dipole/quadrupole H0 anisotropy, and the WALLABY/DESI forecasts. The authors show the dipole is mostly a z-component (0.048 ± 0.014 mag out of 0.063 ± 0.016 mag), i.e., a north-south signal. The paper does a careful job comparing models with BIC, includes the comparison to a bulk flow, and the forecast uses injection-recovery mocks with conservative bulk-flow assumptions. The discussion of sample variance is honest. The direction of the dipole is not aligned with the CMB dipole, which is worth noting.\n\nThe soft spots: Section 3's assumption that variations in a0 are due only to H0, with no photometric calibration differences, is load-bearing. The H I line widths come from heterogeneous catalogs: ALFALFA is all north; southern sources are ADHI, Springob/Cornell, etc. The KS test (p = 0.077) only compares the full line-width distributions; it does not constrain a zero-point offset between north and south. The paper itself notes that Δlog W = 0.01 corresponds to ~3% in H0, the same size as the claimed dipole. No per-source zero-point nuisance parameters are fitted, and no systematic error budget is given. So the 3.9σ is purely statistical. The stress-test concern holds up; this is a real gap.\n\nThat said, the authors are transparent about the bulk-flow degeneracy and even show that the current data can't distinguish a true H0 dipole from a velocity dipole (Bayes factor ~2). The future forecast—1% dipole at 5.8σ with WALLABY/DESI—looks useful and is conservatively constructed. The paper would benefit from quantifying the H I calibration systematic (e.g., by re-fitting with per-catalog zeropoints or adding a nuisance term) and from releasing code.\n\nWho's this for? Cosmologists working on the cosmological principle, H0 tension, or TF peculiar velocities. It deserves serious peer review, but the referee should push on the systematic. My recommendation: send it to review, conditional on the authors addressing the north-south H I calibration concern.","headline":"Solid, honest TF zeropoint dipole analysis whose 3.9σ signal is plausibly a north-south H I calibration artifact; deserves review but needs a systematic error budget.","tokens_in":15322,"tokens_out":2485,"would_cite":true,"duration_ms":17264,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Differential Tully-Fisher measurements across the sky hint at a 3% variation in the Hubble constant, but current data cannot yet tell this apart from a large-scale bulk flow.","keywords":["anisotropic Hubble expansion","Tully-Fisher relation","Cosmicflows-4","cosmological principle","bulk flow","H0 dipole","differential zeropoint"],"falsifier":"Measure the H0 dipole with the combined WALLABY and DESI Tully-Fisher sample: if the recovered dipole amplitude is consistent with zero at the sensitivity that should detect a 1% signal at 5.8σ, the current 3% anisotropy is not a cosmological signal, and the north-south H I line-width heterogeneity (a Δlog W = 0.01 offset) becomes the leading systematic explanation.","tokens_in":14261,"feed_emoji":"🔭","tokens_out":14446,"duration_ms":97274,"temperature":0.7,"pith_summary":"The paper tests the cosmological principle by looking for direction-dependent variation in the Hubble constant, using the Tully-Fisher distance relation applied to the Cosmicflows-4 galaxy catalog. Fitting a dipole to the Tully-Fisher zeropoint yields a 0.063 ± 0.016 mag variation toward (ℓ,b) = (142±30°, 52±10°), which would correspond to a 3% change in H0 (ΔH0 = 2.10 ± 0.53 km/s/Mpc at H0 = 70) at 3.9σ significance. However, a model with a constant H0 plus a ΛCDM-consistent bulk flow is actually favored over the pure H0-dipole interpretation, so the current signal is not yet a detection of anisotropic expansion. The paper's key forward-looking claim is that the combined WALLABY and DESI Tully-Fisher surveys, with an order of magnitude more galaxies and wider redshift coverage, will detect a 1% H0 dipole at 5.8σ and decisively separate it from a bulk flow. This would provide a clean test of whether the Universe expands isotropically at low redshift.","feed_headline":"A 3% H0 dipole is hinted at 3.9σ, but a bulk flow fits too","feed_subtitle":"The signal is degenerate with a bulk flow, but WALLABY and DESI will settle it at 5.8σ.","key_machinery":"The load-bearing object is the Tully-Fisher relation's zeropoint a0, which is degenerate with the Hubble constant via M(w,h) = M(w,h=1) + 5 log h. Because the absolute zeropoint cannot be calibrated, the method uses only differential variations of a0 across the sky, expanding a0(ℓ,b) as a monopole plus dipole (and optionally quadrupole) in Galactic Cartesian components. A measured zeropoint anisotropy Δa0 maps directly to an H0 anisotropy through ΔH0 = H0($10^{{Δa0/5}}$ − 1) (Eq. 3.7), so a dipolar a0 pattern is a dipolar H0 pattern. The analysis is embedded in a forward-modeling fit that simultaneously constrains the Tully-Fisher parameters and the peculiar velocity field, allowing competing models (H0 dipole vs. bulk flow) to be compared through their Bayesian evidence.","core_discovery":"The central claim is that differential measurements of the Tully-Fisher zeropoint across the sky are a clean, calibration-independent way to search for anisotropic Hubble expansion, and that current data already show a hint of such an anisotropy. Using the W1-band Cosmicflows-4 Tully-Fisher sample with cz > 3000 km/s, the authors find a best-fit dipolar zeropoint variation of amplitude 0.063 ± 0.016 mag in the direction (ℓ,b) = (142 ± 30°, 52 ± 10°). Since a zeropoint shift of Δa0 mag corresponds to ΔH0 = H0($10^{{Δa0/5}}$ − 1), this implies ΔH0 = 2.10 ± 0.53 km/s/Mpc at H0 = 70 km/s/Mpc, a 3% sky variation with 3.9σ statistical significance. When the same data are fit with a constant H0 plus a bulk flow, the Bayes factor lnB = 4.7 favors the bulk-flow model over the H0-dipole model, so the anisotropic interpretation is not yet established. Simulations of the upcoming WALLABY and DESI Tully-Fisher datasets show that these surveys, by increasing the sample roughly tenfold and extending to z ≈ 0.1, will detect a 1% H0 dipole at 5.8σ, a 1.2% quadrupole at 5σ, and will clearly distinguish an H0 dipole from a bulk flow, with Bayes factors improving by factors of 25–85.","pith_inferences":["The differential-zeropoint strategy is not limited to Tully-Fisher: applying the same multipole decomposition to a homogeneous all-sky supernova sample that shares a single photometric calibration would provide an independent cross-check of the H0 dipole without the north-south line-width heterogeneity that limits the present analysis.","Because a bulk flow's imprint on the zeropoint scales with redshift while an H0 dipole does not, a single survey that pushes to z > 0.1 with dense sky coverage could separate the two effects even without the full WALLABY+DESI sample size; the redshift lever arm is the physically decisive feature.","If the future surveys confirm the dipole at the 1% level but the direction differs from the current (ℓ,b) = (142°, 52°) maximum, that would point to a systematic in the CF4 H I line-width calibrations rather than a cosmological anisotropy, given how close the current dipole sits to the Zone of Avoidance.","The dominant systematic risk — a north-south offset in H I line-width catalogs as small as Δlog W = 0.01 — can be directly tested by comparing overlapping WALLABY and FASt data in the declination overlap region, providing a near-term empirical check independent of the cosmological interpretation."],"forward_implications":["If the 3% H0 dipole is real, any H0 measurement from a sky region that is not uniformly sampled will carry a direction-dependent bias of up to ~2 km/s/Mpc toward (ℓ,b) ≈ (142°, 52°).","The dipole minimum direction is consistent with the anisotropies reported in galaxy cluster scaling relations, suggesting a common underlying signal that deserves a dedicated comparison across distance indicators.","With WALLABY and DESI data, the method will either confirm a 1% (or larger) H0 dipole at better than 5σ or rule it out, breaking the degeneracy with the local bulk flow.","A detectable quadrupole at the 1.2% level would provide direct evidence for the multipole structure predicted in generalized FLRW frameworks, which current data cannot yet constrain.","Sample variance in the local H0 value (~1.3 km/s/Mpc for the CF4 volume) reduces the significance of the Hubble tension and must be folded into future anisotropy constraints as a systematic floor."],"supporting_citations":[{"why":"Supplies the forward-modeling methodology and the external bulk-flow vector used to generate the mocks and compare models.","marker":"[25]"},{"why":"Provides the Cosmicflows-4 Tully-Fisher sample, the dataset on which all fits are performed.","marker":"[21]"},{"why":"Establishes the direct relation between the Tully-Fisher zeropoint and h = H0/100, the basis for the differential method.","marker":"[17]"},{"why":"Reports an H0 anisotropy from galaxy cluster scaling relations; the TF dipole minimum is compared with this direction.","marker":"[12]"},{"why":"Follow-up analysis of cluster scaling relations that confirms the earlier anisotropy and provides a second comparison direction.","marker":"[13]"},{"why":"Provides the model-independent multipole expansion of luminosity distances that motivates the dipole and quadrupole expansion.","marker":"[14]"},{"why":"Computes expected quadrupole amplitudes in generalized FLRW simulations and applies the formalism to Pantheon+, setting the quadrupole signal the paper tests.","marker":"[15]"},{"why":"Calculates the ΛCDM-expected rms bulk flows at different effective depths, used to set the simulated bulk-flow amplitude for future forecasts.","marker":"[42]"},{"why":"Provides the 2M++ bulk-flow direction used as the input velocity dipole in the mock simulations.","marker":"[43]"}],"fun_headline_variants":["H0 dipole hinted at 3.9σ, but bulk flow may explain it","Anisotropic Hubble expansion: 3.9σ hint, not yet proof","WALLABY and DESI to settle H0 dipole vs bulk flow","3% sky variation in H0: real or bulk flow artifact?"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The analysis assumes that the intrinsic Tully-Fisher relation is everywhere the same and that photometric and H I line-width measurements have no position-dependent calibration differences, so that any angular variation in the fitted zeropoint must be a real variation in H0 (or a real bulk flow).","fun_headline_variants_meta":{"raw":{"variants":["H0 dipole hinted at 3.9σ, but bulk flow may explain it","Anisotropic Hubble expansion: 3.9σ hint, not yet proof","WALLABY and DESI to settle H0 dipole vs bulk flow","3% sky variation in H0: real or bulk flow artifact?"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000931,"raw_usage":{"total_tokens":4108,"prompt_tokens":1189,"completion_tokens":2919,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":805,"completion_tokens_details":{"reasoning_tokens":2835}},"tokens_in":805,"tokens_out":2919,"duration_ms":17682,"temperature":1.0,"reasoning_tokens":2835,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T12:04:40.071060+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the H0 dipole with the combined WALLABY and DESI Tully-Fisher sample: if the recovered dipole amplitude is consistent with zero at the sensitivity that should detect a 1% signal at 5.8σ, the current 3% anisotropy is not a cosmological signal, and the north-south H I line-width heterogeneity (a Δlog W = 0.01 offset) becomes the leading systematic explanation.","supporting_citations":[{"cited_title":"Boubel, M","cited_arxiv_id":null,"evidence_quote":"Supplies the forward-modeling methodology and the external bulk-flow vector used to generate the mocks and compare models."},{"cited_title":"Kourkchi, R.B","cited_arxiv_id":null,"evidence_quote":"Provides the Cosmicflows-4 Tully-Fisher sample, the dataset on which all fits are performed."},{"cited_title":"Boubel, M","cited_arxiv_id":null,"evidence_quote":"Establishes the direct relation between the Tully-Fisher zeropoint and h = H0/100, the basis for the differential method."},{"cited_title":"Migkas, F","cited_arxiv_id":null,"evidence_quote":"Reports an H0 anisotropy from galaxy cluster scaling relations; the TF dipole minimum is compared with this direction."},{"cited_title":"H., Stanford, A., Pacaud, F., Schellenberger, G","cited_arxiv_id":null,"evidence_quote":"Follow-up analysis of cluster scaling relations that confirms the earlier anisotropy and provides a second comparison direction."},{"cited_title":"Heinesen, Multipole decomposition of the general luminosity distance Hubble law — a new framework for observational cosmology , JCAP 2021 (2021) 008","cited_arxiv_id":null,"evidence_quote":"Provides the model-independent multipole expansion of luminosity distances that motivates the dipole and quadrupole expansion."},{"cited_title":"Cowell, S","cited_arxiv_id":null,"evidence_quote":"Computes expected quadrupole amplitudes in generalized FLRW simulations and applies the formalism to Pantheon+, setting the quadrupole signal the paper tests."},{"cited_title":"Whitford, C","cited_arxiv_id":null,"evidence_quote":"Calculates the ΛCDM-expected rms bulk flows at different effective depths, used to set the simulated bulk-flow amplitude for future forecasts."},{"cited_title":"Carrick, S.J","cited_arxiv_id":null,"evidence_quote":"Provides the 2M++ bulk-flow direction used as the input velocity dipole in the mock simulations."}],"review_version":1}